One of the least useful things an ecommerce site can say is:
“No results found.”
Especially when the customer is holding the exact part you sell.
How people actually ask for kart parts
They do not always arrive with a clean SKU.
They say things like:
- “Brake pads for the black Sodi caliper.”
- “The bearing from the rear axle on an SR5.”
- “That pulley, but the 28-tooth one.”
- “It looks like this.”
That is perfectly reasonable. The website should do more of the translation work.
What we changed
RKP search was rebuilt around the way people identify technical parts: SKU where known, but also product names, variants, fitment context and useful descriptive language.
We added a dedicated “Find the right part” experience, predictive search, visible SKUs and variant inventory, and clearer fitment-focused content.
Why this matters commercially
Technical ecommerce has a hidden conversion problem: customers often leave not because the item is unavailable, but because they are not confident they found the right one.
More filters do not automatically solve that. Better context does.
What worked
The strongest improvement has been combining searchable catalogue data with the clues customers already have in their heads.
That also feeds into product content and workshop guides, so search, SEO and support all improve together instead of being separate projects.
The lesson
If customers need to learn your database vocabulary before they can buy from you, the database has won.
I would rather the website understand “the black Sodi one”.
Steal the idea, not the secrets
Want to build your own version?
This is the public build brief: enough architecture to get moving, without the production-specific wiring that took the painful bit to discover.
Build a technical-parts search experience for customers who often do not know the exact SKU.
Requirements:
- Search by SKU, product title, variant, brand, fitment terms and common descriptive language.
- Support useful synonyms and phrases customers naturally use.
- Show SKU and variant availability clearly in results.
- Add predictive search as the customer types.
- Create a dedicated "find the right part" route for ambiguous enquiries.
- Capture zero-result searches so the catalogue can be improved over time.
- Add fitment clues such as dimensions, mounting style, model family or component location.
- Prefer confidence-building context over simply adding more filters.
- Provide a human fallback when the system cannot identify the part confidently.
- Keep searchable product data structured enough to be reused in guides and support tools.
Do not publish internal supplier notes, private margins or uncertain fitment guesses as customer-facing facts.